GPU comparison
AMD Ryzen AI Max+ 395 “Strix Halo” (128GB) vs NVIDIA RTX 5070 Ti (16GB) for local LLMs
How the two stack up for running open-source LLMs locally — memory, bandwidth, price, and how many of the 34 tracked models each one can run.
| AMD Ryzen AI Max+ 395 “Strix Halo” (128GB) | NVIDIA RTX 5070 Ti (16GB) | |
|---|---|---|
| Memory | 128.0 GB | 16.0 GB |
| Bandwidth | 256 GB/s | 896 GB/s |
| Price (approx) | $2,999 | $949 |
| LLMs it runs | 30 of 34 | 18 of 34 |
| Best model it runs | Gemma 4 31B · 10–15 tok/s | Gemma 4 26B-A4B (MoE) · 161–241 tok/s |
The AMD Ryzen AI Max+ 395 “Strix Halo” (128GB) runs 12 more of the tracked models (30 vs 18), thanks to its 128.0 GB of memory. The NVIDIA RTX 5070 Ti (16GB) has more memory bandwidth (896 GB/s), so it generates tokens faster at the same model and quant.